Raising the federal minimum wage to $15 an hour will be hugely beneficial to low-income Americans, according to a new report from the non-partisan Congressional Budget Office.

At the same time, the analysis forecasts the loss of around 1.3 million jobs. There are good reasons not to take this prediction seriously, as I explain below. But even taking CBO’s findings at face value, an overwhelming share of low-wage workers would benefit from the minimum wage increase.

The main takeaway from the CBO’s report is the estimate that a $15 minimum wage by 2025 would raise wages of up to 27.3 million low-wage workers, decrease income inequality, and reduce the number of families in poverty by 1.3 million. Overall, CBO found that low-wage workers as a group would benefit enormously from the minimum wage increase. According to CBO’s own findings, the increase to $15 would raise total annual earnings for low-wage workers by $44 billion.

CBO estimated that up to 27.3 million workers would see wage increases as a result of the policy. About 17.0 million of these low-wage workers would receive direct wage increases because they would otherwise earn less than $15 per hour in 2025, and up to an additional 10.3 million, who would already earn slightly above $15, would also experience some wage increases due to spillover or “ripple” effects.

Taking CBO’s job loss estimates as given implies 95.5 percent of all directly and indirectly affected workers would remain employed and receive a sizable wage increase, and only 4.5 percent would experience job loss. Moreover, as CBO acknowledged, many of those experiencing “job loss” would not be out of work the entire year and may in fact come out even or ahead on an annual basis when they work fewer hours throughout the year but earn higher wages when they do work. Even after accounting for these predicted employment losses, CBO’s findings imply that annual earnings would rise by more than $1,500 for the average directly or indirectly affected low-wage worker.

Just using a conventional cost-benefit framework, it is hard to argue with a policy that reduces poverty and inequality where about 27 million workers would see sizable wage increases and where less than five percent of the affected workforce might see negative effects.

CBO’s job loss estimates are overstated and are inconsistent with CBO’s prior evaluation of minimum wages

The CBO job loss estimate of 1.3 million is much too pessimistic given the highest quality research underlying its analysis. In addition, even though the best recent research is pointing towards smaller employment effects, CBO moved in the opposite direction. In the report just released, CBO assumed even larger employment changes in response to wage increases than it did in its report released five years ago in 2014. To see this, it helps to understand how CBO generated its estimates.

Underlying CBO’s headline numbers are the assumptions it makes about employment effects.

If a $15 minimum wage in 2025 raised the hourly wages of low-wage workers by 10 percent, CBO assumed those workers would be 3.8 percent less likely to be employed. How does that compare to what CBO assumed in 2014 when it evaluated a $10.10 minimum wage increase by 2016? If CBO would have applied its 2014 employment response assumptions to the $15 in 2025 policy, CBO would have predicted about 700 thousand jobs would be lost, as opposed to the 1.3 million it predicted in the new report. In other words, for its latest report, CBO’s new assumptions resulted in an 80% increase in the magnitude of the predicted employment losses for $15 in 2025, relative to what it had predicted back in 2014.

CBO’s change of tune is not justified by the best research published since their last report. One possibility for why CBO assumed more of an employment drag is that because a $15 minimum wage is substantially higher than most previous minimum wages, employment effects might be greater than for smaller increases. This contradicts research CBO cited that looked at much higher state minimum wages and national increases and found little-to-no overall employment change. It’s also at odds with a recent research that CBO did not cite focused on restaurant employment in high minimum wage cities and a just-published study of relatively high state minimum wages in low-wage areas.

In any case, larger employment losses due to higher minimum wage increases can really only explain a small part of CBO’s changing assumptions. Instead, the primary change is that CBO simply increased the magnitude of its employment response assumptions. This is clear from CBO’s new analysis of lower minimum wage targets like $10 and $12 in 2025, where CBO’s underlying assumptions for adult employment effects don’t seem to vary that much across different targets. More concretely, the $10 by 2025 target CBO evaluated in its new report is much lower in inflation-adjusted terms than the proposals it previously evaluated for $10.10 in 2016 or even $9 in 2016. But using the new CBO report’s elasticities for a $10 minimum wage by 2025 generates employment effects about 40 to 100 percent larger than what would be predicted from its 2014 assumptions for $10.10 and $9 in 2016, respectively.

Presumably the major reason for CBO’s larger employment-effect estimates is that it feels the research base now points to even greater disemployment effects. But this is at odds with what even critics of the minimum wage call the “new conventional wisdom,” that minimum wage increases have resulted in little-to-no employment losses. CBO’s central estimates are also much more negative than some of the highest quality recent research I cited above. In particular, my co-authored study with Doruk Cengiz, Arindrajit Dube, and Attila Lindner has the advantages that it concentrates on all low-wage workers and is based on a relatively large number of 138 minimum wage increases. Although CBO incorporated this research, its underlying employment responses are also influenced by other studies that are limited by a small number of minimum wage increases, or those that are limited to particular subgroups of workers like restaurant employees or those with lower education levels.

All told, the main lesson from the CBO report seems correct: the benefits of a gradual increase to a $15 minimum wage vastly outweigh potential downsides. Also correct, in my view, is that in its “likely range” of estimates CBO even acknowledged that a $15 minimum wage by 2025 may cause no employment losses whatsoever. At the same time, its central job loss estimate seems too pessimistic according to my read of the best research and also CBO’s judgment last time around.